Fig. 2.

Comparison of paradigms presented from infant's perspective. Top row shows the screen-based looking tasks; bottom row shows the naturalistic looking task. In each row, the left column shows a photo of the stimulus array from the infant's perspective. The active target for the coding of look durations is indicated using a red rectangle. In the central column the luminance of the images is shown. On the right, feature congestion is shown. Feature congestion quantifies local variability across different first-order features such as colour, orientation and luminance (Rosenholtz et al., 2007) and has been shown to influence gaze allocation in contexts in which no motion is present in the field of view (Henderson and Smith, 2009).